Update to the Reduction API and instances (#476)

* Simplify the macros for declaring and defining the add_device_reduce_instance_xxxx() instances * Change the types of lengths and strides from std::vector to std::array for the reduction device interfaces * Remove DeviceSoftmaxImpl's depending on DeviceReduceMultiblock * Split the cpp and hpp files for reduction instances to enable more parallel compiling * Remove the using of macros for declaring reduction instances and instance references * Update to add_device_reduce_instance_xxxx templated functions * Use ReduceOperation+InElementwiseOp+AccElementwiseOp to repace the ReduceOpId in defining add_reduce_instance_xxxx() templates * Change return format

Update to the Reduction API and instances (#476)
* Simplify the macros for declaring and defining the add_device_reduce_instance_xxxx() instances * Change the types of lengths and strides from std::vector to std::array for the reduction device interfaces * Remove DeviceSoftmaxImpl's depending on DeviceReduceMultiblock * Split the cpp and hpp files for reduction instances to enable more parallel compiling * Remove the using of macros for declaring reduction instances and instance references * Update to add_device_reduce_instance_xxxx templated functions * Use ReduceOperation+InElementwiseOp+AccElementwiseOp to repace the ReduceOpId in defining add_reduce_instance_xxxx() templates * Change return format
dda3a0a1 · Qianfeng · GitHub · 6ea9257e · dda3a0a1 · dda3a0a1
Unverified Commit dda3a0a1 authored Oct 25, 2022 by Qianfeng Committed by GitHub Oct 25, 2022
20 changed files
--- a/library/include/ck/library/utility/host_reduction.hpp
+++ b/library/include/ck/library/utility/host_reduction.hpp
--- a/library/src/tensor_operation_instance/gpu/reduce/CMakeLists.txt
+++ b/library/src/tensor_operation_instance/gpu/reduce/CMakeLists.txt
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_add.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_add.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_amax.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_amax.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_avg.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_avg.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_max.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_max.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_min.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_b16_f32_b16_min.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_amax.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_amax.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_max.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_max.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_min.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f16_f16_min.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16_add.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16_add.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16_avg.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f16_f32_f16_avg.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_i8_i32_i8.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_i8_i32_i8.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32_add.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32_add.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32_amax.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32_amax.cpp
--- a/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32_avg.cpp
+++ b/library/src/tensor_operation_instance/gpu/reduce/device_reduce_instance_blockwise_f32_f32_f32_avg.cpp